M10
Engineering Data Science 1

Back to overview

08:35
conference time (CEST, Berlin)
Application of AI for Pre- and Postprocessing for Crash Simulations
27/10/2021 08:35 conference time (CEST, Berlin)
Room: M
R. Sridhar, P. Kuber, K. Chetti, R. Reddy, N. Kulkarni, M. Mala (Mercedes-Benz Research and Development India Pvt. Ltd., IND)
R. Sridhar, P. Kuber, K. Chetti, R. Reddy, N. Kulkarni, M. Mala (Mercedes-Benz Research and Development India Pvt. Ltd., IND)
Artificial intelligence is one of the most widely and extensively researched topics in the world today. In recent times , AI and machine learning are finding applications in several fields. This paper explores the application of artificial intelligence for pre and post processing , in crash simulations. Abstract: The focus of Artificial Intelligence in Computer Aided Engineering ( CAE ) has largely been on optimization , and prediction methods to replace traditional physics based approaches. While such applications provide a huge advantage with prediction time , they are not suitable for highly non linear behaviour observed in scenarios such as full car crash due to limitations with data , algorithms , computer infrastructure and most of all , since they do not utilize the existing knowledge of physics. Instead of applying Artificial Intelligence in solving , this study aims to improve efficiency by using Artificial Intelligence for pre - and post processing. This paper describes an AI based assistant that uses deep learning, natural language processing ( NLP ) and computer vision ( CV ) to assist with pre- and post processing in Finite Element simulations. In addition to saving the engineer's time on performing repetitive tasks and reducing manual errors , using such an assistant proved to leverage the simplicity of simple human natural language instead of having to train engineers on interfaces of different Computer Aided Engineering tools.
Artificial Intelligence, crash simulations, FEA, FEM, Finite Element, CAE, Non linear FEM, Finite Element Method, NLP, Natural Language Processing
08:55
conference time (CEST, Berlin)
A Flexible and Efficient In-situ Data Analysis Framework for CFD Simulations
27/10/2021 08:55 conference time (CEST, Berlin)
Room: M
C. Gscheidle, J. Garcke (Fraunhofer SCAI, DEU); J. Meng (Hochschule Bonn-Rhein-Sieg, DEU)
C. Gscheidle, J. Garcke (Fraunhofer SCAI, DEU); J. Meng (Hochschule Bonn-Rhein-Sieg, DEU)
The processing of big data from large-scale CFD simulations is a challenges task. On current HPC systems, one observes an increasing gap between the amount of data that is generated during simulations and the amount that can be stored to disk, in particular due to limitations of the data I/O system. For this reason, there is a recent trend for in-situ analyses where the data is processed during the run-time of the simulation and only the analysis results are stored. Furthermore, the extraction of physical and engineering knowledge from highly-resolved simulations leads to a need for new algorithms. While traditional post-processing techniques based on scalar or integral quantities can capture only specific aspects of the flow, the simulation usually contains much more information that can be exploited. Data-driven techniques, e.g. based on Machine Learning (ML), have the potential to extract global features, construct surrogates or find patterns in the data that can be employed in the subsequent scientific and engineering tasks. In this work, we introduce an in-situ data analysis framework with the goal to support both, a flexible application and efficient parallel execution of new algorithms during the development and evaluation process. As most existing Machine Learning libraries provide a Python API and allow an interactive usage through ipython servers, this is also a fundamental design strategy for our software. Therefore, the core API of our framework follows the definition of well-known ML libraries and enhances a simple integration of our code with external methods into a single processing pipeline. By connecting to established in-situ interfaces known from the visualization community, e.g. Catalyst, our code becomes non-intrusive and independent of particular CFD solvers. To meet the goal of an efficient execution, we keep the python layer as thin as possible on the one hand and build on existing math libraries, such as Elemental, for heavy computations on the other hand. To demonstrate the functionality of our framework, a Principal Component Analysis (PCA) of the turbulent velocity field of an industrial relevant use-case, the HVAC duct from an automotive set up, is computed during the simulation based on a batch processing approach. As a result, a significant reduction of data in- and output during and after the simulation can be achieved. A further benefit of the demonstrated in-situ approach lies in earlier availability of the results. This can be exploited by investigating intermediate results during the simulation. A monitoring of the simulation is demonstrated to find early trends in the results by comparing them to previous simulations runs. Based on this, the overall computing time can be reduced by aborting simulations that show non-physical or unwanted behavior.
CFD, HPDA, Machine Learning, in-situ, data analysis
09:15
conference time (CEST, Berlin)
Combining Machine Learning and Simulation for Structural Health Monitoring in Urban Air Mobility
27/10/2021 09:15 conference time (CEST, Berlin)
Room: M
V. Savane (Dassault Systemes, IND); R. Fu (Dassault Systemes, USA)
V. Savane (Dassault Systemes, IND); R. Fu (Dassault Systemes, USA)
For Urban Air Mobility to become a reality, the estimated per passenger cost per mile must decrease by an order of magnitude, to be comparable with the cost for a small sedan. NASA reports that this will be achieved through increased autonomy, technology improvements, and increased operational efficiency. To increase operational efficiency, these vehicles will have to employ Health Utilization Monitoring Systems (HUMS) to help maintenance personnel in identifying and predicting damaged components. HUMS data can add an extra layer of protection against mechanical system failures. In particular, Structural Health Monitoring (SHM) can significantly reduce maintenance and operational costs. One complexity of SHM is in correlating sensor data to damage characteristics. Finite element analysis (FEA) can be combined with machine learning to bridge this gap. This presentation describes a method for detecting cracks in an outer panel based on sensor strain gage data. In this method, diagnostic capabilities are developed based on Artificial Neural Network (ANN) algorithms, and finite element analysis (FEA) results of damaged airframe structure generate the data required for ANN training. Parametric FEM models are employed to consider various crack configurations. The proposed method is applied on a critical zone of an eVTOL fuselage by first creating an FEA model of the fuselage panel, including rivets, stringers, and frames. Next different damage configurations are assessed using the contour integral method and measuring the different stress distributions caused by a progressive crack in critical areas. The FEA approach also suggests the optimal sensor grid design that should be implemented to accurately capture crack characteristics. Sensor outputs, in terms of strains, are measured at different locations and subsequently used in the ANN for predicting crack location and size. Training with the ANN is designed to avoid overfitting on the data and yielded reasonable accuracy in predicting the test data.
Structural Health Monitoring, eVTOL, Simulation, Crack Prediction, Machine Learning
09:35
conference time (CEST, Berlin)
Data Science Meets CFD - How Engineering Can Benefit From Modern Data Science Methods and Techniques
27/10/2021 09:35 conference time (CEST, Berlin)
Room: M
A. Walle (Astrid Walle CFDsolutions, DEU)
A. Walle (Astrid Walle CFDsolutions, DEU)
The combination of Computational Fluid Dynamics (CFD) and Data Science is in the focus of many research projects (Brunton, 2019; Aulich, 2019). Especially the area of physics-informed machine learning aiming to replace CFD at least partially, hence, to save computational costs by inter- and extrapolating fluid fields, attracts much attention. But even before reaching these long-term goals, already today there are a lot of advantages in the wide field of Data Science, which are beneficial for day-to-day engineering tasks and in particular for CFD workstreams. Simulation setups are becoming more complex and memory demanding due to several trends towards more degrees of freedom, towards unsteady analyses, towards parameter space exploration and optimization. This results in a vast number of results and data, which needs to be postprocessed to enable engineers to digest and analyze the data and finally derive insights and gain deeper understanding. Although simulation and postprocessing, hence the data generation, can be automated, there is still manual effort needed to inspect the results to ensure the validity and to understand the underlying mechanisms. These tasks can be supported by methods and techniques from the field of data science. The presented work will show how data access and visualization can be simplified and improved with the help of interactive dashboards. Furthermore, it will be presented how image recognition can support engineers in detecting specific phenomena e.g., separation and how ML can be used to “save” the domain expert’s knowledge. Overall, this talk is an appeal to all engineers to invest some time for data management as this is the required base for all presented activities. Also, the presentation aims to raise awareness for methods and techniques, which can be applied easily and for free and can generate value by enabling faster knowledge transfer. References Brunton, S. L. (2019). Machine Learning for Fluid Mechanics. arXiv. Aulich, M. K. (2019). Surrogate estimations of complete flow fields of fan stage designs via deep neural networks. Proceedings of the ASME Turbo Expo.
Data Science, CFD, Data Management, Interactive Dashboard, Image Recognition
×

[TITLE]

[LISTING

[ABSTRACT]

[DATE]

[ROOM]

[KEYWORDS]